Page MenuHomec4science
Files F61065521

sparse_matrix_petsc.cc
No OneTemporary
Actions

File Metadata

Created
Sat, May 4, 07:51

sparse_matrix_petsc.cc
View Options

/**
* @file sparse_matrix_petsc.cc
*
* @author Aurelia Isabel Cuba Ramos <aurelia.cubaramos@epfl.ch>
*
* @date creation: Mon Dec 13 2010
* @date last modification: Sat Feb 03 2018
*
* @brief Implementation of PETSc matrix class
*
* @section LICENSE
*
* Copyright (©) 2010-2018 EPFL (Ecole Polytechnique Fédérale de Lausanne)
* Laboratory (LSMS - Laboratoire de Simulation en Mécanique des Solides)
*
* Akantu is free software: you can redistribute it and/or modify it under the
* terms of the GNU Lesser General Public License as published by the Free
* Software Foundation, either version 3 of the License, or (at your option) any
* later version.
*
* Akantu is distributed in the hope that it will be useful, but WITHOUT ANY
* WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR
* A PARTICULAR PURPOSE. See the GNU Lesser General Public License for more
* details.
*
* You should have received a copy of the GNU Lesser General Public License
* along with Akantu. If not, see <http://www.gnu.org/licenses/>.
*
*/
/* -------------------------------------------------------------------------- */
#include "sparse_matrix_petsc.hh"
#include "dof_manager_petsc.hh"
#include "mpi_communicator_data.hh"
#include "solver_vector_petsc.hh"
/* -------------------------------------------------------------------------- */
namespace akantu {
/* -------------------------------------------------------------------------- */
SparseMatrixPETSc::SparseMatrixPETSc(DOFManagerPETSc & dof_manager,
const MatrixType & matrix_type,
const ID & id)
: SparseMatrix(dof_manager, matrix_type, id), dof_manager(dof_manager) {
AKANTU_DEBUG_IN();
auto mpi_comm = dof_manager.getMPIComm();
PETSc_call(MatCreate, mpi_comm, &mat);
detail::PETScSetName(mat, id);
resize();
PETSc_call(MatSetFromOptions, mat);
PETSc_call(MatSetUp, mat);
PETSc_call(MatSetOption, mat, MAT_ROW_ORIENTED, PETSC_TRUE);
PETSc_call(MatSetOption, mat, MAT_NEW_NONZERO_LOCATIONS, PETSC_TRUE);
if (matrix_type == _symmetric)
PETSc_call(MatSetOption, mat, MAT_SYMMETRIC, PETSC_TRUE);
AKANTU_DEBUG_OUT();
}
/* -------------------------------------------------------------------------- */
SparseMatrixPETSc::SparseMatrixPETSc(const SparseMatrixPETSc & matrix,
const ID & id)
: SparseMatrix(matrix, id), dof_manager(matrix.dof_manager) {
PETSc_call(MatDuplicate, matrix.mat, MAT_COPY_VALUES, &mat);
detail::PETScSetName(mat, id);
}
/* -------------------------------------------------------------------------- */
SparseMatrixPETSc::~SparseMatrixPETSc() {
AKANTU_DEBUG_IN();
if (mat)
PETSc_call(MatDestroy, &mat);
AKANTU_DEBUG_OUT();
}
/* -------------------------------------------------------------------------- */
void SparseMatrixPETSc::resize() {
auto local_size = dof_manager.getPureLocalSystemSize();
PETSc_call(MatSetSizes, mat, local_size, local_size, size_, size_);
auto & is_ltog_mapping = dof_manager.getISLocalToGlobalMapping();
PETSc_call(MatSetLocalToGlobalMapping, mat, is_ltog_mapping, is_ltog_mapping);
}
/* -------------------------------------------------------------------------- */
/**
* Method to save the nonzero pattern and the values stored at each position
* @param filename name of the file in which the information will be stored
*/
void SparseMatrixPETSc::saveMatrix(const std::string & filename) const {
AKANTU_DEBUG_IN();
auto mpi_comm = dof_manager.getMPIComm();
/// create Petsc viewer
PetscViewer viewer;
PETSc_call(PetscViewerASCIIOpen, mpi_comm, filename.c_str(), &viewer);
PETSc_call(PetscViewerPushFormat, viewer, PETSC_VIEWER_ASCII_MATRIXMARKET);
PETSc_call(MatView, mat, viewer);
PETSc_call(PetscViewerPopFormat, viewer);
PETSc_call(PetscViewerDestroy, &viewer);
AKANTU_DEBUG_OUT();
}
/* -------------------------------------------------------------------------- */
/// Equivalent of *gemv in blas
void SparseMatrixPETSc::matVecMul(const SolverVector & _x, SolverVector & _y,
Real alpha, Real beta) const {
auto & x = aka::as_type<SolverVectorPETSc>(_x);
auto & y = aka::as_type<SolverVectorPETSc>(_y);
// y = alpha A x + beta y
SolverVectorPETSc w(x, this->id + ":tmp");
// w = A x
if (release == 0) {
PETSc_call(VecZeroEntries, w);
} else {
PETSc_call(MatMult, mat, x, w);
}
if (alpha != 1.) {
// w = alpha w
PETSc_call(VecScale, w, alpha);
}
// y = w + beta y
PETSc_call(VecAYPX, y, beta, w);
}
/* -------------------------------------------------------------------------- */
void SparseMatrixPETSc::addMeToImpl(SparseMatrixPETSc & B, Real alpha) const {
PETSc_call(MatAXPY, B.mat, alpha, mat, SAME_NONZERO_PATTERN);
B.release++;
}
/* -------------------------------------------------------------------------- */
/**
* Method to add another PETSc matrix to this PETSc matrix
* @param matrix PETSc matrix to be added
* @param alpha the factor specifying how many times the matrix should be added
*/
void SparseMatrixPETSc::addMeTo(SparseMatrix & B, Real alpha) const {
if (aka::is_of_type<SparseMatrixPETSc>(B)) {
auto & B_petsc = aka::as_type<SparseMatrixPETSc>(B);
this->addMeToImpl(B_petsc, alpha);
} else {
AKANTU_TO_IMPLEMENT();
// this->addMeToTemplated<SparseMatrix>(*this, alpha);
}
}
/* -------------------------------------------------------------------------- */
/**
* MatSetValues() generally caches the values. The matrix is ready to
* use only after MatAssemblyBegin() and MatAssemblyEnd() have been
* called. (http://www.mcs.anl.gov/petsc/)
*/
void SparseMatrixPETSc::applyModifications() {
this->beginAssembly();
this->endAssembly();
}
/* -------------------------------------------------------------------------- */
void SparseMatrixPETSc::beginAssembly() {
PETSc_call(MatAssemblyBegin, mat, MAT_FINAL_ASSEMBLY);
}
/* -------------------------------------------------------------------------- */
void SparseMatrixPETSc::endAssembly() {
PETSc_call(MatAssemblyEnd, mat, MAT_FINAL_ASSEMBLY);
PETSc_call(MatSetOption, mat, MAT_NEW_NONZERO_LOCATIONS, PETSC_FALSE);
this->release++;
}
/* -------------------------------------------------------------------------- */
void SparseMatrixPETSc::copyProfile(const SparseMatrix & other) {
auto & A = aka::as_type<SparseMatrixPETSc>(other);
MatDestroy(&mat);
MatDuplicate(A.mat, MAT_DO_NOT_COPY_VALUES, &mat);
}
/* -------------------------------------------------------------------------- */
void SparseMatrixPETSc::applyBoundary(Real block_val) {
AKANTU_DEBUG_IN();
const auto & blocked_dofs = this->dof_manager.getGlobalBlockedDOFs();
// std::vector<PetscInt> rows;
// for (auto && data : enumerate(blocked)) {
// if (std::get<1>(data)) {
// rows.push_back(std::get<0>(data));
// }
// }
// applyModifications();
static int c = 0;
saveMatrix("before_blocked_" + std::to_string(c) + ".mtx");
PETSc_call(MatZeroRowsColumnsLocal, mat, blocked_dofs.size(),
blocked_dofs.storage(), block_val, nullptr, nullptr);
saveMatrix("after_blocked_" + std::to_string(c) + ".mtx");
++c;
AKANTU_DEBUG_OUT();
}
/* -------------------------------------------------------------------------- */
void SparseMatrixPETSc::mul(Real alpha) {
PETSc_call(MatScale, mat, alpha);
this->release++;
}
/* -------------------------------------------------------------------------- */
void SparseMatrixPETSc::clear() {
PETSc_call(MatZeroEntries, mat);
this->release++;
}
/* -------------------------------------------------------------------------- */
void SparseMatrixPETSc::clearProfile() {
SparseMatrix::clearProfile();
PETSc_call(MatResetPreallocation, mat);
PETSc_call(MatSetOption, mat, MAT_NEW_NONZERO_LOCATIONS, PETSC_TRUE);
// PETSc_call(MatSetOption, MAT_KEEP_NONZERO_PATTERN, PETSC_TRUE);
// PETSc_call(MatSetOption, MAT_NEW_NONZERO_ALLOCATIONS, PETSC_TRUE);
// PETSc_call(MatSetOption, MAT_NEW_NONZERO_ALLOCATION_ERR, PETSC_TRUE);
clear();
}
/* -------------------------------------------------------------------------- */
UInt SparseMatrixPETSc::add(UInt i, UInt j) {
PETSc_call(MatSetValue, mat, i, j, 0, ADD_VALUES);
return 0;
}
/* -------------------------------------------------------------------------- */
void SparseMatrixPETSc::add(UInt i, UInt j, Real val) {
PETSc_call(MatSetValue, mat, i, j, val, ADD_VALUES);
}
/* -------------------------------------------------------------------------- */
void SparseMatrixPETSc::addLocal(UInt i, UInt j) {
PETSc_call(MatSetValueLocal, mat, i, j, 0, ADD_VALUES);
}
/* -------------------------------------------------------------------------- */
void SparseMatrixPETSc::addLocal(UInt i, UInt j, Real val) {
PETSc_call(MatSetValueLocal, mat, i, j, val, ADD_VALUES);
}
/* -------------------------------------------------------------------------- */
void SparseMatrixPETSc::addLocal(const Vector<Int> & rows,
const Vector<Int> & cols,
const Matrix<Real> & vals) {
PETSc_call(MatSetValuesLocal, mat, rows.size(), rows.storage(), cols.size(),
cols.storage(), vals.storage(), ADD_VALUES);
}
/* -------------------------------------------------------------------------- */
void SparseMatrixPETSc::addValues(const Vector<Int> & rows,
const Vector<Int> & cols,
const Matrix<Real> & vals, MatrixType type) {
if (type == _unsymmetric and matrix_type == _symmetric) {
PETSc_call(MatSetOption, mat, MAT_SYMMETRIC, PETSC_FALSE);
PETSc_call(MatSetOption, mat, MAT_STRUCTURALLY_SYMMETRIC, PETSC_FALSE);
}
PETSc_call(MatSetValues, mat, rows.size(), rows.storage(), cols.size(),
cols.storage(), vals.storage(), ADD_VALUES);
}
/* -------------------------------------------------------------------------- */
} // namespace akantu

Event Timeline

c4science · Help